prompt-engineering

Profiles

Engineering profiles for Claude Code — stack-specific conventions, workflows, and automation by otherCode.

What this is

A Claude Code plugin marketplace containing opinionated engineering profiles. Each profile packages the conventions, patterns, and automation for a specific technology stack.

Profiles don't teach Claude what a framework is — it already knows. Instead, they encode your decisions: how you structure code, what patterns to follow, what to avoid, and how to automate enforcement.

Install

Add the marketplace:

/plugin marketplace add othercode/profiles

Install a profile:

/plugin install <profile-name>@othercode

Update

Profiles are distributed as a mono-repo marketplace. To pull the latest versions of all profiles:

/plugin marketplace update othercode

Note: Do not use /plugin update <profile-name> for individually installed profiles — that only works for plugins with external sources (separate GitHub repos). Mono-repo plugins update through the marketplace.

Uninstall

/plugin uninstall <profile-name>@othercode

Available profiles

Profile	Stack	Install
laravel-engineering	PHP, Laravel, Vue.js	/plugin install laravel-engineering@othercode
django-engineering	Python, Django, DRF	/plugin install django-engineering@othercode
prompt-engineering	Claude Code skills, agents, commands	/plugin install prompt-engineering@othercode

Per-project setup

Add this to your project's .claude/settings.json to auto-suggest profiles for anyone who clones the repo:

{
  "extraKnownMarketplaces": {
    "othercode": {
      "source": { "source": "github", "repo": "othercode/profiles" }
    }
  },
  "enabledPlugins": {
    "<profile-name>@othercode": true
  }
}

Philosophy

Each profile is self-contained. No shared base, no cross-profile dependencies. The same concept (e.g. commenting style) may exist in multiple profiles with stack-specific rules and examples — that's intentional.

A profile provides three layers:

• Persona (hook) — sets the engineering mindset on startup
• Skills (auto-triggered) — specific conventions with code examples, activated by context
• Commands (user-invoked) — multi-step workflows triggered explicitly

Creating a new profile

plugins/<name>/
├── .claude-plugin/
│   └── plugin.json
├── commands/
├── agents/
├── skills/
│   └── <skill-name>/
│       └── SKILL.md
└── hooks/
    └── hooks.json

Skills follow the naming convention <verb>-<target> (e.g. writing-tests, commenting-code, implementing-domain-events).

Skill descriptions must be trigger-only — describe when to activate, never summarize what the skill contains.

After creating the plugin structure, add an entry to .claude-plugin/marketplace.json and validate:

/plugin validate .